A novel fusion method based on dynamic threshold neural P systems and nonsubsampled contourlet transform for multi-modality medical images
A novel fusion method based on dynamic threshold neural P systems and nonsubsampled contourlet transform for multi-modality medical images
复制标题
DOI：
10.1016/j.sigpro.2020.107793
复制
发表时间：
2021-01-01
影响因子：
4.4
通讯作者：
Wang, Jun
中科院分区：
文献类型：
--
作者：
Li, Bo;Peng, Hong;Wang, Jun
动态阈值神经P系统（DTNP系统）是一种分布式并行计算模型，其有趣的机制涉及局部区域内神经元的协作发放。在本文中，这一机制相结合的非下采样轮廓波变换（NSCT）开发一种新的融合方法，多模态医学图像。利用改进的新的和-修正拉普拉斯算子（INSML）特征提取多模态图像的互补信息，并将其用于低频NSCT系数的融合规则中。此外，高频NSCT系数提取使用WLE-INSML特征，这是用来构建这些系数的融合规则。所提出的融合方法进行评估的开放数据集组成的12对多模态医学图像。此外，它与9个先前报道的融合方法和4个基于深度学习的融合方法进行了比较。定性和定量的实验结果表明，所提出的融合方法的视觉质量和融合性能的优势。(C)2020 Elsevier B.V.保留所有权利。
Dynamic threshold neural P systems (DTNP systems) are a distributed parallel computing model with an interesting mechanism involving the cooperative spiking of neurons in a local region. In this paper, this mechanism is combined with the nonsubsampled contourlet transform (NSCT) to develop a novel fusion method for multi-modality medical images. The complementary information of multi-modality images is extracted using an improved novel sum-modified Laplacian (INSML) feature, which is used in the fusion rules for the low-frequency NSCT coefficients. Moreover, the high-frequency NSCT coefficients are extracted using the WLE-INSML features, which are used to construct the fusion rules for these coefficients. The proposed fusion method is evaluated on an open dataset consisting of twelve pairs of multi-modality medical images. In addition, it is compared with nine previously reported fusion methods and four deep learning based fusion methods. The qualitative and quantitative experimental results demonstrate the advantage of the proposed fusion method in terms of the visual quality and fusion performance. (C) 2020 Elsevier B.V. All rights reserved.